Project Summary The non-coding RNAs (ncRNAs) play many functional roles in biological processes, such as catalysis, gene expression regulation, and RNA splicing. The various roles played by ncRNAs are determined by their charac- teristic 3D structures. RNA structural motifs are recurrent structural components in ncRNAs. The RNA structural motifs have conserved structures, and therefore, are expected to have similar biological or structural functions. For instance, the kink-turn motif is found in different kinds of ncRNAs and most of them are responsible for protein binding activities. Any alteration of the motif structures has the potential to result in loss-of-function of the RNA structural motif, and in some cases may cause severe diseases. Therefore, the study of RNA structural motif will help us to elucidate the mechanisms of many diseases and lead to the development of novel treatment strategies. Driven by the requirement of biomedical research, many computational methods have been proposed to detect instances of known motifs or to discover novel families. Recently, the rapidly increasing number of resolved RNA 3D structures urges the development of new tools to make use of the huge amount of resources in PDB. In our opinion, the large-scale dataset will support two new approaches for RNA structural motif analysis: generating pro?les to represent motif families and to search for new instances, and identifying the structural and functional relationship among different motifs. In this proposal, we aim at devising a suite of computational methods based on these two ideas. First, we will build 3D position-speci?c scoring matrixes (PSSMs) as pro?les to represent the base-pairing interactions in the discovered RNA structural motif families and develop pro?led-based RNA structural motif search tools. Considering the fact that all the existing RNA structural motif searching programs rely on consensus, the introduction of pro?les will create a great opportunity to improve the search accuracy. In the mean time, the new pro?le model will provide us a formal method to describe motif families, which is very critical for the understanding of the relationship between RNA 3D structures and their functionalities. Second, we will distinguish the structural features of motif family members to identify subfamilies, and study the potential re- lationship among families to recognize clans and modules. The RNA structural motifs from different families may share similar structural features, and they can also interact with each other to perform certain speci?c functions. As far as we know, there is no existing research about the mentioned relationships among RNA structural motifs. Now, the de novo discovery of RNA motif families from the clustering results grants us a solid foundation for further research at both intra- and inter-family levels. With the achievement of these two goals, we also propose to build a new RNA structural motif database to incorporate all the newly discovered knowledge in this project. The tools will also be implemented as user-friendly web servers and integrated with the database to provide a well-rounded service. We expect that the proposed work will lead to better understanding of the RNA structural motifs, and signi?cantly promote biomedical research.